Encoder-Weighted W-Net for Unsupervised Segmentation of Cervix Region in Colposcopy Images

Abstract

Cervical cancer can be prevented and treated better if it is diagnosed early. Colposcopy, a way of clinically looking at the cervix region, is an efficient method for cervical cancer screening and its early detection. The cervix region segmentation significantly affects the performance of computer-aided diagnostics using a colposcopy, particularly cervical intraepithelial neoplasia (CIN) classification. However, there are few studies of cervix segmentation in colposcopy, and no studies of fully unsupervised cervix region detection without image pre- and post-processing. In this study, we propose a deep learning-based unsupervised method to identify cervix regions without pre- and post-processing. A new loss function and a novel scheduling scheme for the baseline W-Net are proposed for fully unsupervised cervix region segmentation in colposcopy. The experimental results showed that the proposed method achieved the best performance in the cervix segmentation with a Dice coefficient of 0.71 with less computational cost. The proposed method produced cervix segmentation masks with more reduction in outliers and can be applied before CIN detection or other diagnoses to improve diagnostic performance. Our results demonstrate that the proposed method not only assists medical specialists in diagnosis in practical situations but also shows the potential of an unsupervised segmentation approach in colposcopy.

Colposcopic Image Segmentation Based on Feature Refinement and Attention

“…Hence, early detection through colposcopy is vital for cervical cancer prevention [2]. However, economic constraints in underdeveloped regions have hindered the widespread adoption of cervical cancer screening [3], resulting in notably higher incidence rates there compared to developed areas [4]- [6]. Introducing Computer-Aided Diagnosis (CAD) technology to in screening could enhance efficiency [7], [8] and extend screening efforts to these underprivileged regions [9], benefiting a broader patient population.…”

“…Most studies focus on the classification of cervical intraepithelial neoplasia (CIN) [11]- [21], with nearly all incorporating a common preprocessing step: extracting ROI from colposcopic examination images. Segmentation is a fundamental step in training deep learning-based models, as more accurate segmentation directly improves the precision of using deep learning for cervical cancer lesion classification [3]. Hence, this study proposes a deep learning-based segmentation model for enhanced segmentation results, aiming to direct clinical doctors' attention to specific areas when analyzing colposcopic images.…”

“…Image segmentation, a critical step in training deep learning-based models, is of utmost importance. More precise segmentation of colposcopic images can greatly improve the accuracy of cervical cancer lesion classification using deep learning techniques[3]. Thus, this study proposes a deep learning-based segmentation model of higher accuracy.…”